APT Testing and 3D Finite Element Analysis of Asphalt Surfacings

1 APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges Dr. M. Huurman1, Assistant Professor Tel: +31 152781525 Email: m.huurman@citg.tudelft.nl T. Medani1, Research Engineer Tel: +31 152781525 Email: t.medani@citg.tudelft.nl Prof.dr. A.A.A. Molenaar1, Professor Tel: +31 152785066 Email: a.a.a.molenaar@citg.tudelft.nl C. Kasbergen2, Research Engineer Tel: +31 152782729 Email: c.kasbergen@citg.tudelft.nl A. Scarpas2, Associate Professor Tel: +31 152784017 Email: a.scarpas@ct.tudelft.nl 1: Section of Road and Railway Engineering, Faculty of Civil Engineering and Geosciences, Delft University of Technology, P.O. Box 5048, 2600 GA Delft, The Netherlands Fax: +31 152783443 Section of Structural Mechanics, Faculty of Civil Engineering and Geosciences, Delft University of Technology, P.O. Box 5048, 2600 GA Delft, The Netherlands 2: Submission date: 18-11-2003 (first submission) 14-05-2004 (after peer review) 4201 words (incl. abstract and references) 3250 words (i.e. 10 figures and 3 tables) 7451 words Word count: Figures and tables: Total: APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 2 ABSTRACT Steel deck plates and surfacings of orthotropic steel deck bridges mostly have a limited life span. To come up with a sound renovation technique for both plate and surfacing the Delft University of Technology is conducting research into bridge deck behaviour. This research considers full scale testing, material testing and Finite Element Modelling (FEM). In this paper attention has been given to the combination of APT testing and FEM. Measurements on a bare bridge panel showed the high accuracy of the structural FEM developed within the research program. Combined with measurements the FEM showed that assumptions on which design methodologies for surfacings are based might not be true. This may explain the short life span in practice. FEM also showed large effects of surfacing stiffness, membrane stiffness and bridge geometry on bridge deck and material responses. Because measurements cannot be performed for all possible combinations, FEM acts to generalise the measurement data and thus enlarges their value. The accuracy of the FEM is limited when surfacing and interface materials are considered to respond elastically. It is concluded that the accuracy increases when the complex non-linear response of the surfacing and the membrane interface is considered. For this reason material research is currently being conducted. It is expected that general insight into bridge deck behaviour is obtained in the near future, when the results of full scale APT measurements, FEM and material research can be combined. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 3 1. INTRODUCTION Modern steel deck bridges consist of a 10-14 mm thick steel deck plate stiffened by U-shaped longitudinal stiffeners. To support the deck crossbeams are welded to the structure every few meters, figure 1. Usually, the deck plate is surfaced with a 30-70 mm thick surfacing material (1). A mix of coarse aggregate, sand and filler with high bitumen content is often used as surfacing material, i.e. mastic asphalt. In practice the surfacing of orthotrophic steel deck bridges tends to show a relatively short lifespan. Furthermore the steel deck plate proves to be very vulnerable to fatigue related damage and cracks in the steel deck plate are common problems. Orthotropic steel decks are often used in larger bridges crossing the main waterways of the Netherlands. These bridges are part of the primary road network. Resurfacing and/or maintenance of the deck plate on these bridges either results in a strong reduction of road capacity or long detours. The Dutch public does not accept either. To come up with renovation techniques that will guarantee a long lifespan of both the deck plate and the surfacing, a research program is currently being executed at the Delft University of Technology. Three sections of the department of Materials, Mechanics and Structures (MMS) are involved. The section of Steel Structures is doing research into the development of fatigue damage in the steel components of the bridge. The section of Road and Railway Engineering and the section of Structural Mechanics are responsible for research into surfacing behaviour, which is the topic of this paper. To the best of the authors’ knowledge there is no universally accepted methodology for the design of surfacings on orthotropic steel deck bridges. The design is merely based on experience and some norms obtained from structural tests e.g. in the Netherlands this might be the fracture energy obtained from Semi-Circular Bending tests (2). However, some theories to estimate the stresses/strains in the different layers are available, e.g. (3, 4, 5, 6, 7). Almost all researchers adopted one or both of the following assumptions: - Linear strain gradient in the surfacing and the steel. - The gradient of strain through the depth of the surfacing and steel are equal. However, there is no theoretical and/or experimental background for such assumptions. Hameau et al. (8) executed an experimental program on a two-span beam model. The model was tested using sinusoidal loading. The measured strain over the height of the asphalt surfacing and the steel is shown in figure 2. The figure clearly shows that the strain gradient in the surfacing is not linear. This non-linearity may be a result of the non-linear response of the asphalt surfacing and/or the geometry of the tested model. However, the work indicates that the assumptions upon which most composite theories are based might not be true. This suggests that the traditional design techniques may well be too crude for the proper design of bridge surfacings and steel deck plates. Combined with an increase in traffic loads in terms of numbers, wheel load magnitudes and contact pressures that are beyond our experience, this may explain why prevailing design methods have very limited success. The above indicates the need for better insight into the behaviour of orthotropic steel deck bridge surfacings and the interaction hereof with the steel structure of the bridge. This insight should lead to a design procedure that enables the design of surfacings such that a longer lifespan is guarantied. Furthermore the procedure should be able to determine the structural value of the surfacing, so that steel deck plate fatigue damage may be prevented by proper surfacing design. To obtain the required insight the research concentrates on three main issues, i.e.: material testing, full scale testing and structural modelling. In this paper a 3D-FEM (Finite Element Model) of the bridge deck and the surfacing will be presented. The model is verified on the basis of full-scale measurements with the LINTRACK APT facility. It is explained at the end of the paper that accurate material modelling will result in a better APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 4 understanding of the structural distress phenomena and the parameters that influence them. Therefore material testing is currently being executed at the Delft University of Technology. 2. FULL SCALE TESTING 2.1 The LINTRACK APT facility LINTRACK is a facility for full-scale Accelerated Pavement Testing. LINTRACK is located at the outdoor test area of the Road and Railway Research Laboratory of the Delft University of Technology. The linear LINTRACK facility primarily consists of a dual steel gantry, along which a loading carriage can move forward and backward, figure 3. A single dual or wide base truck wheel can be mounted in the loading carriage, which allows the wheel load to be adjusted from 15 to 100 kN (9, 10). The maximum speed of the loading carriage is 20 km/h, but lower speeds are also possible. Over a length of 4 m the speed of the loading carriage is constant. About 500 forward and 500 backward wheel movements per hour can be accomplished. LINTRACK allows for the application of lateral wander, up to 1 m to either side of the wheel track centre line. The LINTRACK APT facility is placed in a shelter. Infrared radiators were implemented in 1997. This enables control of the pavement temperature during testing, up to 30ºC to 35ºC (depending on the wind speed) above ambient temperature. 2.2 LINTRACK APT measurements For full-scale experiments in the LINTRACK facility two orthotropic steel deck bridge test panels are available. These are standard orthotropic steel decks with a thin steel deck plate of 10 mm. The distance between the crossbeams in the test panels is 2 m, though in practice a larger spacing is common. The crossbeams are continuously supported in LINTRACK, while in real bridges they are only supported at the main girders of a bridge. These are two significant differences between the test panels and real bridges. The test panels have a width of 4 m, which allows for the definition of two, 2 m wide, tests sections per panel. Measurements on the first test panel have recently been completed. At the start of the test program benchmark measurements were performed on both sections of the panel. These benchmark measurements considered the bare bridge panel, figure 3. After the benchmark tests surfacing materials were applied to the panel. Section 1 was surfaced with a 50 mm thick layer of conventional mastic asphalt and section 2 with a 50 mm thick layer of open graded polymer bound aggregate. Thereafter measurements on the two sections with surfacing were performed. The test panel response was monitored by approximately 150 strain gauges. First results of the measurements, which have been carried out by the section of Steel Structures, are reported extensively in (11). These measurements were carried out on both unsurfaced test sections and also on section 1 after surfacing with conventional mastic asphalt at a temperature of 0oC. Four wheel-configurations were used to load the bridge deck: - Single wheel (Type A Eurocode 1-3, 220 x 320 mm2) - Double wheel (Type B Eurocode 1-3, 2 x 220 x 320 mm2) - Supper Single wheel (Type C Eurocode 1-3, 270 x 320 mm2) - Super Super Single wheel (Type D, 500 x 300 mm2) Measurements were performed at load magnitudes of 25 and 50 kN and at speeds of 2 and 20 km/h. Due to space limitations, it is not possible to discuss all measured data. The data however showed that maximum strains in both the surfacing and the steel deck plate are often measured in the gauges that measure transversal strain at the centreline of a longitudinal stiffener between crossbeams. The locations of these gauges S09, D09 and S53 are indicated in figure 1. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 5 An example of the measurement results is given in figure 4. This figure shows the strain measured by gauge S09. The plotted strains are measured under a 50 kN load and plotted as a function of the wheel position for both a type A and type B wheel configuration. 3. FINITE ELEMENT MODELLING The LINTRACK measurements enable engineers to have an accurate view of bridge response in reality. However valuable, this view lacks perspective. The measurements briefly discussed earlier for instance only consider one type of bridge geometry in combination with two types of surfacing. As a result only limited insight is obtained into the influence of important structural issues on the strain development in the deck plate and the surfacing. Examples of these issues are: - the membrane interface material response (type of material and temperature), - the membrane interface thickness, - the deck plate thickness, - the surfacing thickness, - real bridge supports, - girder spacing, - the surfacing material response (type of material and temperature). Combined with a structural model the value of the measurements strongly increases as a result. For this reason two structural FEM's of the bridge deck were developed within the (Computer Aided Pavement Analysis) CAPA FE-package. The models make use of symmetry in a vertical plane perpendicular to the direction of the bridge span, see figure 5. Due to the use of symmetry the models enable cost effective determination of stresses and strains in the bridge panel. A disadvantage is that the models are only applicable for loads at the plane of symmetry. One model may be used to compute stress and strain in the case that the load is placed over a crossbeam, while the other model is valid in the case that the load is placed between two crossbeams. The models do not result in any restrictions with respect to the load location in lateral direction. The element meshes of the two models are very similar and based on the same basic mesh. The basics of the meshes are discussed elsewhere (12). 4. MODEL VERIFICATION It is not possible to make computations for all individual measurements. Accurate calculations are only possible when there is synergy between the load case geometry and individual elements within the mesh. In effect this implies that a separate mesh is required for each individual wheel load configuration and each wheel load position. It was decided arbitrarily that the model should be validated on the measurement data obtained with gauge S09. This gauge was located at the upper surface of the deck plate between two crossbeams over the centreline of a longitudinal stiffener. The measurements showed that the maximum strain in the steel deck plate often occurs at the location of strain gauge S09. 4.1 Bare steel deck Figure 6 gives an example of the modelled response of the unsurfaced bridge deck. Table 1 gives an indication of the agreement between measured and simulated bridge response. The observed differences remain within a range of -6.6% to 6.9%. Measurements were performed on the two test sections in the same panel. Due to symmetry these measurements should lead to the same strain for the unsurfaced sections. However, this is not the case. The same holds for the wheel load speed. Measurements are performed at 2 km/h and 20 km/h. Since steel does not have frequency dependant behaviour the measurements should be independent of speed, however, differences in the measurements are observed. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 6 From the differences in the various measurements it was determined that the measurements show a scatter of about 6%. For the measurements with gauge S09 this is shown in table 2. The scatter in the data is thus of the same order of magnitude as the differences between the calculated and the measured strains. This shows the very high accuracy of the finite element model. 4.2 Deck with mastic asphalt surfacing From the point of view of a road engineer, steel is a material that shows an almost ideal mechanical behaviour. When steel is not stressed to its plastic limit it can accurately be modelled linear elastically. Since the steel of the bridge deck was not stressed to its plastic limit the verification of the model on the basis of the bridge deck without surfacing really comes down to the verification of the finite element mesh. This verification thus showed that the modelled bridge geometry in combination with the subdivision hereof in individual elements leads to an accurate structural model. When the bridge is surfaced the situation becomes much more complex. Now the results of the calculation will also depend on the mechanical behaviour of the surfacing material and the membrane interface bonding the surfacing to the steel deck plate. The membrane interface is a 3 mm thick bituminous-based layer. Due to the high bitumen content both the membrane interface and the surfacing material will show a complex mechanical response which is dependent on frequency (strain rate) and temperature. The response will show elastic, viscous and plastic components, as is the case for any bitumen bound material. The properties of the membrane and the surfacing are not known yet and these materials are therefore modelled linear elastically now. In any mechanical problem the development of stresses and strains is dependent on the load, the geometry and the material behaviour. An example of the deformed mesh with an indication of strain is given in figure 7. This figure gives a good impression of the complex field of deformation that is a result of the complex 3D geometry. It is thus expected that the geometry of the bridge deck is a very important factor in explaining stresses and strains. To verify this statement use is made of the measurements by Hameau, figure 2. This measurement shows non-linear development of strain over the height of the surfacing. Figure 8 shows the development of lateral strain over the height of the surfacing as calculated with the model. By comparing figures 2 and 8 it is shown that the phenomenon measured by Hameau is properly reflected in the model. Figure 8 very clearly indicates that the strain gradient in the surfacing is not linear. Furthermore it also indicates that the strain gradient in the surfacing is not necessarily equal to the gradient in the steel. It is thus concluded that both assumptions on which current design methods are based might not be true. LINTRACK measurements on the bridge surfaced with mastic asphalt are only available for the type C loading. The temperature of the mastic asphalt during the measurements was 0oC. It is these measurements that have been back-calculated. In total 10 calculations have been made. Table 3 shows that the strain measured at the surface of the mastic asphalt surfacing is back-calculated quite accurately without as long as the combined bending stiffness of the steel deck, the membrane interface and the mastic asphalt surfacing remains within certain realistic limits. From the table it is concluded that accurate (error <10%) results are obtained as long as the stiffness of the mastic asphalt does not exceed 9000 MPa. Such a high surfacing stiffness should be compensated by a marginal membrane interface shear stiffness. With a decrease of the surfacing stiffness to 6500 MPa accurate surfacing strains are only obtained when the stiffness of the membrane interface increases to 20 N/mm3. It is stated that the range of surfacing stiffness at which accurate surfacing strains are obtained is realistic considering the low temperature of 0oC. Back-calculating the measured strain in the steel deck plate proved to be more troublesome. Table 3 shows that there is no combination of surfacing stiffness and membrane interface stiffness that will lead to a proper back-calculated strain in the steel and in the surfacing. The best estimate of the APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 7 strain in the steel deck in combination with an accurate surfacing strain, is obtained when a stiff surfacing is placed over an extremely soft membrane interface. 5. FULL 3D MODEL FOR MOVING LOADS The LINTRACK measurements considered one bridge geometry in combination with two types of surfacing materials and membrane interfaces. However, the previous sections indicate that the strains (and stresses) that develop in both the steel and the surfacing are, apart from the loading, strongly dependent on bridge geometry, surfacing material response and membrane interface behaviour. It is because of these interactions, complete insight into the behaviour of orthotropic steel deck bridges with surfacings can only be obtained when the measurements are generalised by theoretical modelling. Calculations on the bridge without a surfacing showed the high accuracy of the structural model. The accuracy of the model, however, remains limited when considering a surfaced bridge deck and not taking into account the complex non-linear behaviour of the membrane and the surfacing. The strong effects of the membrane stiffness and the surfacing stiffness on calculated strain indicate the importance of material behaviour on the total bridge deck response. It is thus expected that the model accuracy will strongly improve when the response of both the membrane interface and the surfacing material are more realistically modelled. A material model capable in describing the complex response of bitumen bound materials is incorporated in CAPA. This Asphalt Concrete REsponse (ACRE) model is extensively discussed elsewhere (13, 14, 15) and considers the rate of strain as a very important parameter in explaining material behaviour. This implies that the combination of the structural model and the incorporated material model will only lead to proper results when the strain development in the calculations is similar to the strain development in reality. In the case of the measurements on the orthotropic steel deck bridge this implies that a moving wheel has to be considered. For this purpose a full 3D model is required. Such a model was thus built, the mesh of the model is shown in figure 9. The full 3D-model is costly in terms of computational power and thus only serves research purposes. As a preparation measure a simulation with the full 3D model was performed. Since it is expected that bridge dynamics may play a role in the real bridge deck response this is a dynamic calculation simulating the bridge response under a moving load while taking into account non-linear visco-elasto-plastic material response. The calculation is based on surfacing material parameters retrieved from other research on mastic asphalt. The properties of the membrane were chosen arbitrarily. In figure 10 the development of damage in the surfacing under three load cycles is presented. The figure indicates that the full 3D-model is operative. The figure furthermore shows that maximum material damage is to be expected between crossbeams. In the full 3D-model crossbeams were placed at 3 m. The difference between the damage under the two individual tyres of the double wheel indicates that the support of the crossbeam to the deck plate influences the deck response even in the middle of two crossbeams. In the measurements the crossbeams are placed at 2 m, resulting in even higher effects of crossbeam support. 6. FUTURE WORK The previous section shows that a full 3D model in which the complex response of the surfacing and the membrane interface are considered is available. It was also shown that the bridge response is apart from bridge geometry and loading strongly dependant on membrane interface and surfacing material response. For this reason laboratory material research into the behaviour of the membrane interface and the surfacing is currently being executed. Dynamic and monotonic shear tests are being executed on the membrane interface material. These tests are designed to give insight into the membrane response (stiffness) and the membrane strength (failure envelop). On the surfacing materials: dynamic and monotonic uniaxial tension and compression tests are executed. These tests will give insight into the response and the failure envelopes of several surfacing materials. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 8 When the material tests are completed further calculations into the bridge deck behaviour are planned. It is expected that these calculations will show a higher accuracy than the linear calculations discussed earlier in this paper. With this being the case the much-needed general insight into bridge deck behaviour could be obtained. 7. CONCLUSIONS On the basis of the material presented in this paper the following conclusions can be drawn: 1 From practise it is known that surfacings on orthotropic steel deck bridges have a limited life span and fatigue cracking is a common type of damage in the steel deck plates. 2 The assumptions upon which current design techniques are based proved not to be true. 3 Full-scale LINTRACK measurements were done on a single bridge geometry that does not fully represent the geometry of real bridge decks in combination with two surfacing materials. 4 The results of the measurements become of much greater value when supported by structural modelling. 5 The high accuracy of the geometrical models is shown by verification on the basis of the unsurfaced steel deck. 6 Calculations have shown that strain gradients are strongly influenced by the bridge deck geometry. 7 Calculations were made on the basis of highly simplified membrane interface and surfacing behaviour, i.e. adaptation of linear elastic material response. These showed that it is relatively easy to accurately back-calculate strains in the surfacing. However, it is not possible to accurately back-calculate the strain in the deck plate of the steel deck bridge. 8 Linear elastic calculations showed that the development of strain throughout the bridge including surfacing is, apart from geometry and loading, strongly determined by the behaviour of both the membrane interface and the surfacing material. 9 It is expected that the accuracy of the structural model will improve when the response of the interface and the surfacing material are considered. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 9 REFERENCES 1 Kolstein, M.H. and Wardenier, J., Stress Reduction due to Surfacing on Orthotropic Steel Decks, Proceedings ISAB Workshop, Evaluation of Existing Steel and Composite Bridges, Laussanne, 1997. 2 NPC, Research into Mastic Asphalt Mix of Ewijk Bridge (in Dutch), Utrecht, the Netherlands, 1996. 3 Metcalf, C.T., Flexural Tests of paving Materials for Orthotropic Steel Plate Bridges, Highway Research Record No. 155, Washington, D.C, 1967. 4 Sedlacek, G., and Bild, St., Research into the Durability of Wearing Courses (in German), Herausgegeben vom Bunddesminister für Verkehr, 1985. 5 Kolstein, M.H., Economic Optimization of Thick Wearing and Isolation Courses on Steel Bridge Decks, Criteria and Factors (in Dutch), Report 26.6.90.12/A2/22.03, Stevin Laboratory, Faculty of Civil Engineering, Delft, 1990. 6 Cullimore, M.S.G., Flett, I.D., Smith, J.W., Flexure of Steel Bridge Deck Plate with Asphalt Surfacing, IABSE Periodical 1/1983, University of Bristol, pp. 58-83, 1983. 7 Nakanishi, N. and Okochi, T., The Structural Evaluation for an Asphalt Pavement on a Steel Plate Deck, Proceedings of the First International Conference, World of Asphalt Pavement (AAPA), Sydney, Australia pp. 112-123, 2000. 8 Hameau, G., Puch, C., Ajour, A.M., Fatigue Behaviour due to Negative Bending Moments (in French), Revétments de Chaussées sùr platelages métalliques, 1981. 9 Groenendijk, J., Accelerated testing and surface cracking of asphaltic concrete pavements, PhD dissertation, Delft University of Technology, Delft, 1998. 10 Houben, L.J.M, Kooij, J. van der, Naus, R.W.M. and Bhairo, P.D., APT Testing of Modular Pavement Structure ‘Easy Road’ and Comparison with conventional asphalt motorway structures, APT 2004. 11 Jong, de F.B.P., Kolstein, M.H. and Bijlaard, F.S.K. Development of Long Term Renovation Techniques for Orthotropic Steel Bridge Decks with Fatigue Cracks, Proceedings of the 3rd European Conference on Steel Structures, 19-20 September 2002, Coimbra, Portugal APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 10 12 Huurman, M., Medani, T.O , Molenaar, A.A.A, Kasbergen, C., Liu, X., and Scarpas, A. 3D-FEM for the estimation of the behaviour of asphaltic surfacings on orthotropic steel deck bridges, 3rd International Symposium on 3D Finite Element for Pavement Analysis, Design & Research, Amsterdam, The Netherlands, 2002 13 Scarpas, A., Al-Khoury, R.I.N., Gurp, C.A.P.M. van and Erkens, S.M.J.G., Finite Elements Simulation of Damage Development in Asphalt Concrete Pavements, 8th International Conference on Asphalt Pavements (ICAP), University of Washington, Seattle, U.S.A, 1997. 14 Erkens, S.M.J.G., Liu, X. and Scarpas, A., 3D Finite Element Model for Asphalt Concrete Response Simulation, The International Journal of Geomechanics, Volume 2, Number 3 305-330, 2002. 15 Erkens, S.M.J.G ., Asphalt Concrete Response (ACRe) - determination, modelling and prediction-, PhD dissertation, Delft University of Technology, Delft, 2002. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 11 LIST OF FIGURES FIGURE 1 Example of the geometry of an orthotropic steel deck bridge. FIGURE 2 Strain distribution (Hameau, 8). FIGURE 3 LINTRACK loading carriage on the test panel without surfacing. FIGURE 4 Measured strains between two girders and over a stiffener as a function of the wheel position (left: Wheel A, 50 kN / right: Wheel B, 50 kN), (de Jong et.al. 11). FIGURE 5 CAPA models that use symmetry, for loads between crossbeams (left) and on crossbeams (right). FIGURE 6 Indication of lateral strain εxx in the deformed bridge deck without surfacing (150x) under a 50 kN dual wheel. FIGURE 7 Deformed bridge deck (250x) with transversal strain εxx under a 14.5 tons dual wheel axle (50 mm 5000 MPa surfacing on 1 N/mm3 membrane interface). FIGURE 8 Calculated lateral strain over the hight. FIGURE 9 a) top view of the full 3D-model, b) bottom view of the full 3D-model. FIGURE 10 Total damage under the first three passages of a 14.5 ton dual wheel axle. LIST OF TABLES TABLE 1 A Comparison between the measured and simulated strain TABLE 2 Scatter in the measurement TABLE 3 A Comparison between the measured and simulated strain using different membrane stiffness APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 12 Asphaltic surfacing S53 300 mm Binder layer Steel deck plate (12 mm) D09 S09 325 mm 535 mm 200 mm Cross section at crossbeam 300 mm Steel stiffener (6 mm) Spacing in girder Crossbeam Crossbeam flange (18 mm) Transversal cross section FIGURE 1 Example of the geometry of an orthotropic steel deck bridge. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 13 FIGURE 2 Strain distribution (Hameau, 8). APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 14 FIGURE 3 LINTRACK loading carriage on the test panel without surfacing. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 15 FIGURE 4 Measured strains between two girders and over a stiffener as a function of the wheel position (left: Wheel A, 50 kN / right: Wheel B, 50 kN), (de Jong et.al. 11). APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 16 FIGURE 5 CAPA models that use symmetry, for loads between crossbeams (left) and on crossbeams (right). APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 17 Transversal strain εxx [µ-strain]: -250 0.0 500 FIGURE 6 Indication of lateral strain εxx in the deformed bridge deck without surfacing (150x) under a 50 kN dual wheel. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 18 Transversal strain εxx [µ-strain]: -850 0.0 900 FIGURE 7 Deformed bridge deck (250x) with transversal strain εxx under a 14.5 tons dual wheel axle (50 mm 5000 MPa surfacing on 1 N/mm3 membrane interface). APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 19 εxx under a 14.5 tons dual wheel single axle between girders 50 mm surfacing, surfacing stiffness = 3000 MPa, Cross beam spacing = 3000 mm ε xx [µ-s train] -400 0 10 20 Z [m ] m 30 40 50 -300 -200 -100 0 100 200 300 400 section 1 section: 1 2 3 4 5 3 membrane shear stiffness: 1, 10, 100, 10000 N/mm ε xx [µ-s train] -700 -600 -500 -400 -300 -200 -100 0 10 20 Z [m ] m 0 membrane shear stiffness ε xx [µ-s train] -750 -500 -250 0 250 500 750 1000 1250 60 100 200 300 400 500 600 700 -1250 -1000 0 section 2 Z [m ] m 10 20 30 40 50 60 section 3 30 40 50 60 membrane shear stiffness ε xx [µ-s train] -800 -600 -400 -200 0 200 400 600 800 1000 membrane shear stiffness ε xx [µ-s train] -1000 0 10 20 Z [m ] m 30 40 50 60 -500 0 10 20 Z [m ] m 30 40 50 60 -400 -300 -200 -100 0 100 200 300 400 section 4 section 5 membrane shear stiffness membrane shear stiffness FIGURE 8 Calculated lateral strain over the hight. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 20 FIGURE 9 a) top view of the full 3D-model, b) bottom view of the full 3D-model. APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 21 Repetition Repetition Repetition 55 FIGURE 10 Total damage ε [µ-strain]: 500 Total damage under the first three passages of a 14.5 ton dual wheel APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 22 TABLE 1 Strain Gauge S09 S09 S09 S09 Wheel Load Type A B C C A Comparison between the measured and simulated strain Measured Simulated Load Speed Difference Strain Strain [kN] 50 50 50 25 [km/h] 2 2 2 2 [µm/m] -695 -408 -636 -416 [µm/m] -708 -381 -680 -399 % 1.9 -6.6 6.9 -4.1 APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 23 TABLE 2 Strain Gauge S09 Speed S09 S09 S09 Symmetry S09 S53 Scatter in the measurement Wheel Load Type C C C C C C Load [kN] 50 50 25 25 50 50 Speed [km/h] 02 20 02 20 02 02 Measured Strain [µm/m] -636 -645 -416 -441 -636 -640 Difference [%] 1.4 6.0 0.6 APT Testing and 3D Finite Element Analysis of Asphalt Surfacings on Orthotropic Steel Deck Bridges 24 TABLE 3 Smix [MPa] A Comparison between the measured and simulated strain using different membrane stiffness Membrane shear Strain at steel deck surface, Strain at mastic asphalt surface, i.e. gauge S09 i.e. gauge D09 stiffness [N/mm3] (measured –165 µm/m) (measured -641 µm/m) difference difference [µm/m] [µm/m] 20 10 20 5 1 0.3 0.2 0.05 0.4 0.25 -63 -88 -51 -101 -131 -130 -123 -124 -119 -120 -61.8% -46.7 % -69.1 % -38.8 % -20.6 % -21.2 % -25.4% -24.8 % -27.8 % -27.3 % -635 -692 -577 -676 -735 -697 -667 -670 -656 -653 -0.94 % 7.95 % -9.98 % 5.46 % 14.66 % 8.73 % 4.06 % 4.52 % 2.34 % 1.87 % 6500 6500 7500 7500 7500 8200 8200 8200 9000 9000